capacitor-plugin-upgrade-v6-to-v7
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFEDYNAMIC_CONTEXT_INJECTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_CONTEXT_INJECTION]: The skill employs the
!command`` syntax inSKILL.mdto execute shell commands when the skill is loaded. These commands (node -eandfind) are used to gather a 'Live Project Snapshot' of the current repository state. - [COMMAND_EXECUTION]: Shell commands are executed at load time via dynamic context to inspect project files. Specifically: a Node.js one-liner reads and parses
package.jsonto extract Capacitor-related dependencies, and thefindcommand locates specific directories likeiosandandroid. - [INDIRECT_PROMPT_INJECTION]: The skill ingests data from the project's
package.jsonfile through the dynamic snapshot. This represents a technical surface for indirect prompt injection. 1. Ingestion points: Readspackage.jsonvia a load-time Node.js script inSKILL.md. 2. Boundary markers: None explicitly defined for the snapshot output. 3. Capability inventory: The agent is authorized to useBashfor restricted commands and is directed to perform file edits and runnpm install. 4. Sanitization: The extraction script filters entries to only include those starting with@capacitor/.
Audit Metadata